The present invention relates to an integrated circuit having a compensation component with a reduced and adjustable on resistance. This compensation component may be a vertical compensation component or alternatively a lateral compensation component. In one embodiment, the compensation component is a power field effect transistor.
With respect to compensation components, as is known, there is an extremely extensive prior art concerned with the formation of the compensation regions. In this respect, as examples and representative of further documents, reference shall be made in particular to the following:
U.S. Pat. No. 6,630,698 B1 illustrates a field effect transistor in which compensation regions in the form of p-conducting pillars have a variable doping, so that the p-conducting pillars have a higher doping in a region near the source than in a region near the drain. In this case, the p-conducting pillars always have the same, constant cross-sectional area.
In U.S. Pat. No. 6,639,272 B2, a likewise variable doping in p-conducting pillars of compensation regions is achieved by means of differing layer thickness of individual epitaxial layers in conjunction with the p-conducting pillars having a cross-sectional area that is essentially identical over their length.
Further examples of compensation components comprising compensation regions having a constant cross-sectional area are given by U.S. Pat. Nos. 4,754,310 A1, 5,216,275 A1, 6,621,122 B2 and US 2004/108568 A1.
Furthermore, compensation components using trench technology are known for example from U.S. Pat. Nos. 6,512,267 B2, 6,410,958 B1 and 6,433,385 B1. In these documents, too, the compensation regions have a largely constant cross-sectional area over their length. Only in U.S. Pat. No. 6,433,385 B1 mentioned last is a description given of a trench transistor having an “extended p-zone” (extended p-conducting zone) which is embedded between oxide-filled trenches and acts as a compensation region, here the trench having a smaller cross-sectional area in its lower section than in its upper section, so that the compensation region has a larger cross-sectional area in deeper regions of the drift zone than in less deep regions.
U.S. Pat. No. 6,677,643 B2 discloses a compensation component in which compensation regions have a larger pitch in proximity to the source than in proximity to the drain, whereby structures can arise in which a compensation region having a larger cross-sectional area adjoins a compensation region having a smaller cross-sectional area in a vertical direction between source electrode and drain electrode.
While the conventional compensation components previously mentioned above all have a vertical structure, U.S. Pat. No. 6,858,884 B2 describes a compensation component with a lateral structure, here a compensation region decreasing in terms of its cross-sectional area in the direction between the source electrode and the drain electrode. However, the compensation region extends directly on the drain side as far as a highly doped substrate, so that no “pedestal layer” remains between the substrate and the compensation region. This document does not discuss any relationships between the form of the compensation region and the capacitance profile of the compensation component as a function of the drain-source voltage.
Finally, WO 2005/065385 A2 discloses a compensation component in the form of a field effect transistor in which floating p-conducting compensation regions lie in the drift zone between source and drain, the diameters of the compensation regions decreasing with increasing distance from the source electrode. The aim of reducing the diameter is to produce floating p-conducting regions that are intended to achieve an increased breakdown voltage. Continuous compensation pillars are not provided in the case of this compensation component.
The numerous documents above are cited by way of example for the extensive prior art with respect to compensation components, to which attention has already been drawn. It must be emphasized, however, that in these documents at any rate and also in the other prior art investigated, there is no explicit discussion of the relationship between the form of the compensation regions, that is to say the geometrical shape thereof, and the profile of the capacitance of the compensation component as a function of the voltage present between drain and source, that is to say the output capacitance.
Investigations have now illustrated that especially high-voltage power transistors embodied as compensation components have the particular property that in them the output capacitance is very large given small drain-source voltages, but decreases rapidly by several orders of magnitude as the drain-source voltage increases, the transition between the range with high output capacitance and the range with no output capacitance not in any way being effected in continuous fashion, but rather being effected in stepped fashion.
The above dependence of the output capacitance Coss on the drain-source voltage VDS is qualitatively illustrated schematically in FIG. 1 in a logarithmic representation. The individual steps with which the output capacitance of an investigated compensation field effect transistor falls rapidly as the drain-source voltage VDS increases can clearly be discerned here.
The physical background for this rapid fall in the output capacitance will be explained in more detail below. It should be noted here that the feedback capacitance, that is to say the capacitance between gate and drain, behaves in similar fashion, and it assumes even smaller values than the output capacitance on account of the drain-source capacitance additionally contained in the output capacitance.
FIG. 2 schematically illustrates a p-conducting compensation region 5 in an n-conducting drift zone 2, the pn junction 10 between the compensation region 5 and the drift zone 2 being in pillar-type form. The source contact lies at the upper edge of FIG. 2, while the drain contact is to be assumed at the lower edge.
If 10 V are present between the drain contact and the source contact in the switched-on state of the compensation component, then a space charge zone 9 forms which extends all around the pillar-type compensation region 5 if the interior of the compensation region remains at source potential, but the area surrounding the compensation region 5 in the drift zone 2 rises slowly up to the drain potential on account of the bulk resistance in the drift zone 2. The space charge zone 9 is particularly extended at the lower, drain-side end of the compensation region, so that especially here only a relatively narrowly delimited region remains for the current flow in the drift zone 2.
If, then, a plurality of compensation regions lie with their longitudinal extent parallel to one another between a source contact and a drain contact and the voltage present between drain and source is increased continuously, the space charge zones 9 of the parallel compensation regions will finally converge. At the instant when the space charge zones 9 converge, the geometry and the effective thickness of the space charge zone change significantly, which is manifested in a step in the capacitance profile. The steep fall in the output capacitance (cf. FIG. 1) and its stepped profile are therefore ultimately caused by the convergence of the space charge zones, in which case it should be taken into consideration that the space charge zones first touch one another at the drain-side, lower end of the compensation regions. At this moment when they touch, the entire upper part of the drift path becomes ineffective for the capacitance, whereby the pronounced steps can be explained.
The steep profile of the output capacitance according to the example of FIG. 1 results in steep voltage edges, which are highly unfavorable for the electromagnetic compatibility (EMC behavior) of a circuit that uses such a compensation component. This holds true in particular for the range of large steps in the profile of the output capacitance.
To summarize, therefore, it can be established that a less stepped profile of the output capacitance as a function of the drain-source voltage would be highly favorable for the EMC behavior of a compensation component.
For these and other reasons, there is a need for the present invention.